Background: Our objective was to examine the accuracy of intracardiac ultrasound (ICUS) measurement of left ventricular (LV) volumes and ejection fraction (EF) using a 10-MHz ultrasound catheter. ICUS can image the LV in cross sections at all levels along the long axis with a transducer mounted on the tip of a catheter. Sequential serial LV cross-sectional images can be obtained during cardiac catheterization and used to calculate LV volumes by Simpson's rule. This technique may be an alternative to contrast LV angiography.
Methods and results: A beating-heart in vivo model was created to measure LV volume directly and continuously with an intracavity high-compliance latex balloon connected to a calibrated extracardiac reservoir in eight dogs in 35 experimental stages. A 10F ICUS catheter with a 10-MHz single-element transducer was introduced retrogradely via the aortic valve to the apex. Series of sequential LV cross-sectional images were recorded from the apex to the base during a calibrated pullback of the catheter. At each 5-mm interval, the LV cross section was traced at end diastole and end systole. LV volume was calculated by Simpson's rule by integrating all segmental areas multiplied by segmental height. The effect on accuracy of selecting 5-, 10-, or 15-mm heights or a single section at the midventricular level for measurement was assessed. The influence of distorted ventricular shape on the accuracy of ICUS measurements of LV volume was evaluated. This method was applied in 19 experimental stages in 10 intact dogs and pigs catheterized via the femoral artery. In the in vivo canine model, LV end-diastolic volume, end-systolic volume, and EF determined by ICUS using 5-, 10-, or 15-mm segments were not different from the actual measurements. But correlation and agreement between ICUS end-diastolic volume and direct measurements for 5- and 10-mm segments were significantly better than for 15-mm segments or a single section. Similar excellent correlations and agreement were observed for actual and ICUS-derived end-systolic volumes using 5-, 10-, or 15-mm segments. The ICUS-derived EF correlated very well with actual EF with a small measurement error of 3.91 +/- 2.59% for 5-mm or 4.13 +/- 2.79% for 10-mm segments but a significantly greater measurement error for 15-mm segments (5.35 +/- 3.76%) or single sections (14.8 +/- 12.2%). The presence of LV infarction or aneurysm did not significantly influence the accuracy of ICUS calculations for segmental heights < or = 10 mm. Application in intact animals demonstrated a good correlation between stroke volume measured by ICUS and by thermodilution or flowmeter. ICUS-derived LV volumes correlated well with biplane angiographic volumes, with a tendency toward underestimation. There was no significant difference between ICUS-determined LV EF and EF determined by angiography.
Conclusions: Intracardiac echocardiography accurately measures LV volumes and global systolic function in both regularly shaped and distorted left ventricles. This technique directly and continuously visualizes circumferential LV endocardium and wall thickness without contrast agents or geometric assumptions for calculation of LV volume. Thus, it should be particularly useful in patients at high risk for contrast-related complications or distorted LV shapes in which geometric assumptions may not be valid.